The Gemini 2.5 model features a 1 million-token long context window, allowing users to include entire folders in prompts for more relevant responses.

On June 12, 2025, Google Cloud announced an upgrade to Gemini Code Assist, enhancing its AI-driven coding capabilities with the introduction of the Gemini 2.5 model.

Developers can now customize the assistant's behavior with new features such as custom commands for reusable prompts and project-level Rules to adhere to internal coding standards.

Code Assist can follow user-defined 'Rules' to ensure that the generated code aligns with best practices and established libraries.

A new tabbed chat functionality enables users to manage multiple coding problems simultaneously while retaining chat history.

Users can initiate new chat sessions easily by clicking the New Chat (+) button, allowing for multitasking without losing conversation history.

The Quick Pick menu in VS Code and JetBrains IDE allows users to create shortcuts for repetitive tasks, streamlining their coding workflow.

The upgrade includes a pull-request reviewer that identifies logic gaps in code diffs before they are reviewed by human teammates.

The new model significantly improves the assistant's ability to suggest better code, address complex questions and bugs, and analyze code changes more thoroughly.

Individuals can access a free tier, while organizations benefit from policy controls and an enhanced review agent through Gemini for Google Cloud, with optional GitHub integration.

Gemini 2.5 Pro is available in AI Studio and the Gemini Advanced subscription, with a forthcoming 'Deep Think' mode planned to enhance reasoning capabilities for large codebases and complex mathematics.

An internal study revealed that teams utilizing Gemini Code Assist were 2.5 times more likely to complete standard coding tasks.